test_flock/test_lockf are racy

[andreas-schwab]

The tests test_flock and test_lockf are racy and fail spuriously. The parent process trying to synchronize the subprocess may sleep up to max_delay=1.0 seconds between tries, while the subprocess locks the file at most sleep_time=0.2 seconds until it exits (and unlocks the file). This may happen completely while the parent sleeps between two tries.

Linked PRs

• gh-107851: reduce race condition in test_flock and test_lockf #107931
• gh-107851: Fix spurious failures in fcntl eintr tests #121556
• [3.13] gh-107851: Fix spurious failures in fcntl eintr tests (GH-121556) #121585
• [3.12] gh-107851: Fix spurious failures in fcntl eintr tests (GH-121556) #121586

[Eclips4]

Hello!
Can you give an example of the failure of these tests?
Did you see failure at our buildbot fleet?

[andreas-schwab]

You get a timeout while waiting for the synchronisation.

[colesbury]

I've been seeing this fairly often in the nogil refleaks buildbot (even before we disabled the GIL in the CI).

For example, https://buildbot.python.org/all/#/builders/1287/builds/218/steps/5/logs/stdio

======================================================================
ERROR: test_flock (__main__.FNTLEINTRTest.test_flock)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/ubuntu/buildarea/pull_request.itamaro-ubuntu-aws.refleak.nogil/build/Lib/test/_test_eintr.py", line 535, in test_flock
    self._lock(fcntl.flock, "flock")
    ~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ubuntu/buildarea/pull_request.itamaro-ubuntu-aws.refleak.nogil/build/Lib/test/_test_eintr.py", line 517, in _lock
    raise Exception("failed to sync child in %.1f sec" % dt)
Exception: failed to sync child in 300.4 sec
======================================================================
ERROR: test_lockf (__main__.FNTLEINTRTest.test_lockf)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/ubuntu/buildarea/pull_request.itamaro-ubuntu-aws.refleak.nogil/build/Lib/test/_test_eintr.py", line 532, in test_lockf
    self._lock(fcntl.lockf, "lockf")
    ~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^
  File "/home/ubuntu/buildarea/pull_request.itamaro-ubuntu-aws.refleak.nogil/build/Lib/test/_test_eintr.py", line 517, in _lock
    raise Exception("failed to sync child in %.1f sec" % dt)
Exception: failed to sync child in 300.6 sec